Image processing in welding seam tracking based on single-stripe laser

The image processing methods in welding seam tracking based on single-stripe laser are discussed.The wavelet transform is applied firstly,and the horizontal detail part of the image is gotten;less information of noise is embraced in this part,while the information of laser stripe is well preserved.Taking this part for subsequent seam recognition,its remaining noise is eliminated by using the median filtering in the vertical direction.Finally,KSW entropy algorithm is used for image segmentation.Good effects can be gotten using these methods,and the anti-disturbance performance is well,and the time cost is little enough to meet the requirement of real-time tracking.